[Clang][AVX512][BUILTIN] Adding intrinsics for VSCATTERPF{1|0}{DPS|QPS|DPD|QPD} instruction set

[igorb]

please update diff after http://reviews.llvm.org/D19224 committed.

lib/Headers/avx512fintrin.h

6324 ↗

(On Diff #54331)

most probably this one is from http://reviews.llvm.org/D19224 , please remove it.

[igorb]

LGTM

cfe/trunk/lib/Headers/avx512fintrin.h

__builtin_ia32_gathersiv8di ((__v8di) _mm512_undefined_epi32(),\
                               __addr, (__v8si) __index, (__mmask8) -1 , __scale);\
 })
 
 #define _mm512_mask_i32gather_epi64( __v1_old,  __mask, __index, __addr, __scale) __extension__ ({\
 __builtin_ia32_gathersiv8di ((__v8di) __v1_old,\
                               __addr, (__v8si) __index, __mask , __scale);\
 })
 
+#define _mm512_i64scatter_ps(__addr,__index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scatterdiv16sf(__addr, (__mmask8) -1,\
+                             (__v8di) __index, (__v8sf) __v1, __scale);\
+})
+
+#define _mm512_mask_i64scatter_ps(__addr, __mask,__index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scatterdiv16sf(__addr, __mask,\
+                             (__v8di) __index, (__v8sf) __v1, __scale);\
+})
+
+#define _mm512_i64scatter_epi32(__addr, __index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scatterdiv16si (__addr, (__mmask8) -1,\
+                              (__v8di) __index, (__v8si) __v1, __scale);\
+})
+
+#define _mm512_mask_i64scatter_epi32(__addr, __mask, __index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scatterdiv16si (__addr, __mask, (__v8di) __index,\
+                              (__v8si) __v1, __scale);\
+})
+
+#define _mm512_i64scatter_pd( __addr, __index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scatterdiv8df (__addr, (__mmask8) -1,\
+                             (__v8di) __index, (__v8df) __v1, __scale);\
+})
+
+#define _mm512_mask_i64scatter_pd( __addr, __mask, __index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scatterdiv8df (__addr, __mask, (__v8di) __index,\
+                             (__v8df) __v1, __scale);\
+})
+
+#define _mm512_i64scatter_epi64( __addr, __index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scatterdiv8di (__addr, (__mmask8) -1,\
+                             (__v8di) __index, (__v8di) __v1, __scale);\
+})
+
+#define _mm512_mask_i64scatter_epi64( __addr, __mask, __index,  __v1,  __scale) __extension__ ({\
+__builtin_ia32_scatterdiv8di(__addr, __mask, (__v8di) __index,\
+                            (__v8di) __v1, __scale);\
+})
+
+#define _mm512_i32scatter_ps( __addr, __index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scattersiv16sf (__addr, (__mmask16) -1,\
+                              (__v16si) __index, (__v16sf) __v1, __scale);\
+})
+
+#define _mm512_mask_i32scatter_ps( __addr, __mask, __index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scattersiv16sf (__addr, __mask, (__v16si) __index,\
+                              (__v16sf) __v1, __scale);\
+})
+
+#define _mm512_i32scatter_epi32( __addr, __index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scattersiv16si (__addr, (__mmask16) -1,\
+                              (__v16si) __index, (__v16si) __v1, __scale);\
+})
+
+#define _mm512_mask_i32scatter_epi32( __addr, __mask, __index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scattersiv16si (__addr, __mask, (__v16si) __index,\
+                              (__v16si) __v1, __scale);\
+})
+
+#define _mm512_i32scatter_pd( __addr, __index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scattersiv8df (__addr, (__mmask8) -1,\
+                             (__v8si) __index, (__v8df) __v1, __scale);\
+})
+
+#define _mm512_mask_i32scatter_pd( __addr, __mask, __index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scattersiv8df (__addr, __mask, (__v8si) __index,\
+                             (__v8df) __v1, __scale);\
+})
+
+#define _mm512_i32scatter_epi64( __addr, __index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scattersiv8di (__addr, (__mmask8) -1,\
+                             (__v8si) __index, (__v8di) __v1, __scale);\
+})
+
+#define _mm512_mask_i32scatter_epi64( __addr, __mask, __index, __v1, __scale) __extension__ ({\
+__builtin_ia32_scattersiv8di (__addr, __mask, (__v8si) __index,\
+                             (__v8di) __v1, __scale);\
+})
+
 static __inline__ __m128 __DEFAULT_FN_ATTRS
 _mm_mask_fmadd_ss (__m128 __W, __mmask8 __U, __m128 __A, __m128 __B) 
 {
  return (__m128) __builtin_ia32_vfmaddss3_mask ((__v4sf) __A,
           (__v4sf) __B,
           (__v4sf) __W,
           (__mmask8) __U,
           _MM_FROUND_CUR_DIRECTION);

cfe/trunk/lib/Headers/avx512pfintrin.h

 #define _mm512_mask_prefetch_i64gather_pd( index,  mask, addr,  scale, hint) __extension__ ({\
 __builtin_ia32_gatherpfqpd (mask, (__v8di) index, (long long const *) addr, scale, hint);\
 })
 
 #define _mm512_mask_prefetch_i64gather_ps( index,  mask, addr, scale,  hint) ({\
 __builtin_ia32_gatherpfqps (mask, (__v8di) index, (int const *) addr, scale, hint);\
 })
 
+#define _mm512_prefetch_i32scatter_pd(addr,  index,  scale,  hint) __extension__ ({\
+__builtin_ia32_scatterpfdpd ((__mmask8) -1, (__v8si) index, \
+                            (void  *)addr, scale, hint);\
+})
+
+#define _mm512_mask_prefetch_i32scatter_pd(addr,  mask,  index,  scale,  hint) __extension__ ({\
+__builtin_ia32_scatterpfdpd (mask, (__v8si) index, (void  *) addr,\
+                             scale, hint);\
+})
+
+#define _mm512_prefetch_i32scatter_ps(addr, index, scale, hint) __extension__ ({\
+__builtin_ia32_scatterpfdps ((__mmask16) -1, (__v16si) index, (void  *) addr,\
+                             scale, hint);\
+})
+
+#define _mm512_mask_prefetch_i32scatter_ps(addr, mask, index, scale, hint) __extension__ ({\
+__builtin_ia32_scatterpfdps (mask, (__v16si) index, (void  *) addr,\
+                             scale, hint);\
+})
+
+#define _mm512_prefetch_i64scatter_pd(addr, index, scale, hint) __extension__ ({\
+__builtin_ia32_scatterpfqpd ((__mmask8) -1, (__v8di) index, (void  *) addr,\
+                             scale, hint);\
+})
+
+#define _mm512_mask_prefetch_i64scatter_pd(addr, mask, index, scale, hint) __extension__ ({\
+__builtin_ia32_scatterpfqpd (mask, (__v8di) index, (void  *) addr,\
+                             scale, hint);\
+})
+
+#define _mm512_prefetch_i64scatter_ps(addr, index, scale, hint) __extension__ ({\
+__builtin_ia32_scatterpfqps ((__mmask8) -1, (__v8di) index, (void  *) addr,\
+                             scale, hint);\
+})
+
+#define _mm512_mask_prefetch_i64scatter_ps(addr, mask, index, scale, hint) __extension__ ({\
+__builtin_ia32_scatterpfqps (mask, (__v8di) index, (void  *) addr,\
+                             scale, hint);\
+})
 
 #undef __DEFAULT_FN_ATTRS
 
 #endif

cfe/trunk/test/CodeGen/avx512f-builtins.c

 }
 
 __m512i test_mm512_mask_i32gather_epi64(__m512i __v1_old, __mmask8 __mask, __m256i __index, void const *__addr) {
   // CHECK-LABEL: @test_mm512_mask_i32gather_epi64
   // CHECK: @llvm.x86.avx512.gather.dpq.512
   return _mm512_mask_i32gather_epi64(__v1_old, __mask, __index, __addr, 2); 
 }
 
+void test_mm512_i64scatter_ps(void *__addr, __m512i __index, __m256 __v1) {
+  // CHECK-LABEL: @test_mm512_i64scatter_ps
+  // CHECK: @llvm.x86.avx512.scatter.qps.512
+  return _mm512_i64scatter_ps(__addr, __index, __v1, 2); 
+}
+
+void test_mm512_mask_i64scatter_ps(void *__addr, __mmask8 __mask, __m512i __index, __m256 __v1) {
+  // CHECK-LABEL: @test_mm512_mask_i64scatter_ps
+  // CHECK: @llvm.x86.avx512.scatter.qps.512
+  return _mm512_mask_i64scatter_ps(__addr, __mask, __index, __v1, 2); 
+}
+
+void test_mm512_i64scatter_epi32(void *__addr, __m512i __index, __m256i __v1) {
+  // CHECK-LABEL: @test_mm512_i64scatter_epi32
+  // CHECK: @llvm.x86.avx512.scatter.qpi.512
+  return _mm512_i64scatter_epi32(__addr, __index, __v1, 2); 
+}
+
+void test_mm512_mask_i64scatter_epi32(void *__addr, __mmask8 __mask, __m512i __index, __m256i __v1) {
+  // CHECK-LABEL: @test_mm512_mask_i64scatter_epi32
+  // CHECK: @llvm.x86.avx512.scatter.qpi.512
+  return _mm512_mask_i64scatter_epi32(__addr, __mask, __index, __v1, 2); 
+}
+
+void test_mm512_i64scatter_pd(void *__addr, __m512i __index, __m512d __v1) {
+  // CHECK-LABEL: @test_mm512_i64scatter_pd
+  // CHECK: @llvm.x86.avx512.scatter.qpd.512
+  return _mm512_i64scatter_pd(__addr, __index, __v1, 2); 
+}
+
+void test_mm512_mask_i64scatter_pd(void *__addr, __mmask8 __mask, __m512i __index, __m512d __v1) {
+  // CHECK-LABEL: @test_mm512_mask_i64scatter_pd
+  // CHECK: @llvm.x86.avx512.scatter.qpd.512
+  return _mm512_mask_i64scatter_pd(__addr, __mask, __index, __v1, 2); 
+}
+
+void test_mm512_i64scatter_epi64(void *__addr, __m512i __index, __m512i __v1) {
+  // CHECK-LABEL: @test_mm512_i64scatter_epi64
+  // CHECK: @llvm.x86.avx512.scatter.qpq.512
+  return _mm512_i64scatter_epi64(__addr, __index, __v1, 2); 
+}
+
+void test_mm512_mask_i64scatter_epi64(void *__addr, __mmask8 __mask, __m512i __index, __m512i __v1) {
+  // CHECK-LABEL: @test_mm512_mask_i64scatter_epi64
+  // CHECK: @llvm.x86.avx512.scatter.qpq.512
+  return _mm512_mask_i64scatter_epi64(__addr, __mask, __index, __v1, 2); 
+}
+
+void test_mm512_i32scatter_ps(void *__addr, __m512i __index, __m512 __v1) {
+  // CHECK-LABEL: @test_mm512_i32scatter_ps
+  // CHECK: @llvm.x86.avx512.scatter.dps.512
+  return _mm512_i32scatter_ps(__addr, __index, __v1, 2); 
+}
+
+void test_mm512_mask_i32scatter_ps(void *__addr, __mmask16 __mask, __m512i __index, __m512 __v1) {
+  // CHECK-LABEL: @test_mm512_mask_i32scatter_ps
+  // CHECK: @llvm.x86.avx512.scatter.dps.512
+  return _mm512_mask_i32scatter_ps(__addr, __mask, __index, __v1, 2); 
+}
+
+void test_mm512_i32scatter_epi32(void *__addr, __m512i __index, __m512i __v1) {
+  // CHECK-LABEL: @test_mm512_i32scatter_epi32
+  // CHECK: @llvm.x86.avx512.scatter.dpi.512
+  return _mm512_i32scatter_epi32(__addr, __index, __v1, 2); 
+}
+
+void test_mm512_mask_i32scatter_epi32(void *__addr, __mmask16 __mask, __m512i __index, __m512i __v1) {
+  // CHECK-LABEL: @test_mm512_mask_i32scatter_epi32
+  // CHECK: @llvm.x86.avx512.scatter.dpi.512
+  return _mm512_mask_i32scatter_epi32(__addr, __mask, __index, __v1, 2); 
+}
+
+void test_mm512_i32scatter_pd(void *__addr, __m256i __index, __m512d __v1) {
+  // CHECK-LABEL: @test_mm512_i32scatter_pd
+  // CHECK: @llvm.x86.avx512.scatter.dpd.512
+  return _mm512_i32scatter_pd(__addr, __index, __v1, 2); 
+}
+
+void test_mm512_mask_i32scatter_pd(void *__addr, __mmask8 __mask, __m256i __index, __m512d __v1) {
+  // CHECK-LABEL: @test_mm512_mask_i32scatter_pd
+  // CHECK: @llvm.x86.avx512.scatter.dpd.512
+  return _mm512_mask_i32scatter_pd(__addr, __mask, __index, __v1, 2); 
+}
+
+void test_mm512_i32scatter_epi64(void *__addr, __m256i __index, __m512i __v1) {
+  // CHECK-LABEL: @test_mm512_i32scatter_epi64
+  // CHECK: @llvm.x86.avx512.scatter.dpq.512
+  return _mm512_i32scatter_epi64(__addr, __index, __v1, 2); 
+}
+
+void test_mm512_mask_i32scatter_epi64(void *__addr, __mmask8 __mask, __m256i __index, __m512i __v1) {
+  // CHECK-LABEL: @test_mm512_mask_i32scatter_epi64
+  // CHECK: @llvm.x86.avx512.scatter.dpq.512
+  return _mm512_mask_i32scatter_epi64(__addr, __mask, __index, __v1, 2); 
+}
+
 __m128d test_mm_mask_rsqrt14_sd(__m128d __W, __mmask8 __U, __m128d __A, __m128d __B){
   // CHECK-LABEL: @test_mm_mask_rsqrt14_sd
   // CHECK: @llvm.x86.avx512.rsqrt14.sd
   return _mm_mask_rsqrt14_sd(__W, __U, __A, __B);
 }
 
 __m128d test_mm_maskz_rsqrt14_sd(__mmask8 __U, __m128d __A, __m128d __B){
   // CHECK-LABEL: @test_mm_maskz_rsqrt14_sd

cfe/trunk/test/CodeGen/avx512pf-builtins.c

void test_mm512_mask_prefetch_i64gather_pd(__m512i index, __mmask8 mask, void const *addr, int hint) {
   return _mm512_mask_prefetch_i64gather_pd(index, mask, addr, 2, 1); 
 }
 
 void test_mm512_mask_prefetch_i64gather_ps(__m512i index, __mmask8 mask, void const *addr, int hint) {
   // CHECK-LABEL: @test_mm512_mask_prefetch_i64gather_ps
   // CHECK: @llvm.x86.avx512.gatherpf.qps
   return _mm512_mask_prefetch_i64gather_ps(index, mask, addr, 2, 1); 
 }
+
+void test_mm512_prefetch_i32scatter_pd(void *addr, __m256i index) {
+  // CHECK-LABEL: @test_mm512_prefetch_i32scatter_pd
+  // CHECK: @llvm.x86.avx512.scatterpf.dpd.512
+  return _mm512_prefetch_i32scatter_pd(addr, index, 1, 2); 
+}
+
+void test_mm512_mask_prefetch_i32scatter_pd(void *addr, __mmask8 mask, __m256i index) {
+  // CHECK-LABEL: @test_mm512_mask_prefetch_i32scatter_pd
+  // CHECK: @llvm.x86.avx512.scatterpf.dpd.512
+  return _mm512_mask_prefetch_i32scatter_pd(addr, mask, index, 1, 2); 
+}
+
+void test_mm512_prefetch_i32scatter_ps(void *addr, __m512i index) {
+  // CHECK-LABEL: @test_mm512_prefetch_i32scatter_ps
+  // CHECK: @llvm.x86.avx512.scatterpf.dps.512
+  return _mm512_prefetch_i32scatter_ps(addr, index, 1, 2); 
+}
+
+void test_mm512_mask_prefetch_i32scatter_ps(void *addr, __mmask16 mask, __m512i index) {
+  // CHECK-LABEL: @test_mm512_mask_prefetch_i32scatter_ps
+  // CHECK: @llvm.x86.avx512.scatterpf.dps.512
+  return _mm512_mask_prefetch_i32scatter_ps(addr, mask, index, 1, 2); 
+}
+
+void test_mm512_prefetch_i64scatter_pd(void *addr, __m512i index) {
+  // CHECK-LABEL: @test_mm512_prefetch_i64scatter_pd
+  // CHECK: @llvm.x86.avx512.scatterpf.qpd.512
+  return _mm512_prefetch_i64scatter_pd(addr, index, 1, 2); 
+}
+
+void test_mm512_mask_prefetch_i64scatter_pd(void *addr, __mmask16 mask, __m512i index) {
+  // CHECK-LABEL: @test_mm512_mask_prefetch_i64scatter_pd
+  // CHECK: @llvm.x86.avx512.scatterpf.qpd.512
+  return _mm512_mask_prefetch_i64scatter_pd(addr, mask, index, 1, 2); 
+}
+
+void test_mm512_prefetch_i64scatter_ps(void *addr, __m512i index) {
+  // CHECK-LABEL: @test_mm512_prefetch_i64scatter_ps
+  // CHECK: @llvm.x86.avx512.scatterpf.qps.512
+  return _mm512_prefetch_i64scatter_ps(addr, index, 1, 2); 
+}
+
+void test_mm512_mask_prefetch_i64scatter_ps(void *addr, __mmask16 mask, __m512i index) {
+  // CHECK-LABEL: @test_mm512_mask_prefetch_i64scatter_ps
+  // CHECK: @llvm.x86.avx512.scatterpf.qps.512
+  return _mm512_mask_prefetch_i64scatter_ps(addr, mask, index, 1, 2); 
+}